22. Errors I. Proofreading and MMR (1,519; 8/10)

DNA polymerases are astonishingly accurate--but not perfect. Let me quantify that.

The leading and lagging strand DNA polymerases, which are the workhorses of genome replication, initially make a nucleotide addition error once every 10,000 to 100,000 nucleotides. That’s impressive, but not nearly accurate enough.

That level of mis-incorporation would result in tens to hundreds of thousands of errors, or mutations, in a replicated human genome. And that, in turn, could cause a disastrous increase in the incidence of potential cancer cells.

Fortunately, we known that when a human genome is normally replicated, it contains very few errors--zero to a handful. So something else must be going on. That "something else" is DNA repair. And, in fact, many human diseases, especially cancers, are linked to problems in DNA repair.

The two repair solutions we're about to discuss, DNA polymerase proofreading and a multi-protein pathway called mismatch repair (MMR) reduce the number of errors to close to zero. Proofreading is performed by the polymerase itself when it senses that it has made an error. Proofreading reduces errors by roughly 100- to 1,000-fold. MMR delivers an additional significant improvement that can also reach hundreds- to roughly a thousand-fold.

These two repair solutions brings the number of errors in a replicated human genome down to what scientists have observed: a very small number of errors, or even no errors.

You might ask why we're devoting time to DNA repair in a book about DNA replication? Because the two are inseparable. Replication inevitably generates errors. Those errors must be fixed. Also, as we'll soon see, DNA lesions caused by factors unrelated to replication also occur in large numbers. Those must be fixed, too.

Replication and repair are so tightly linked that we can't ignore the topic. And we wouldn’t want to. DNA repair is among the most remarkable abilities of cells.

Types of Errors

Before describing these two error correction solutions, let me clarify what I mean by an error. For the moment, we'll consider three kinds of errors in the strand being synthesized: substitutions, insertions, and deletions

Substitutions create nucleotide mismatches. The DNA polymerase simply adds the wrong nucleotide. It inserts, for example, an "A" nucleotide across from a parental template strand "C" instead of the correct "G". Substitutions are like misspellings.

The other two kinds of errors are insertions and deletions. An insertion is the addition of an extra, unpaired nucleotide on the new strand being synthesized. A deletion is the failure to add a paired nucleotide--one that ordinarily should be there--on the new strand.

Here's an example that's closer to home. A letter "t" insertion in the word "nucleotide" might result in the misspelled word "nucleottide" with that extra unwanted "t." A letter "t" deletion might result in the misspelling "nucleoide" with no "t."

Insertions and deletions often occur when a DNA polymerase slips when copying a repetitive sequence--a long stretch of As, for example. The enzyme in a sense loses track of where it is on the parental template strand.

With an insertion, the extra nucleotide on the new strand doesn't pair with any nucleotide on the template strand. So it generates a small loop or protrusion on the new strand. With a deletion error, the unpaired nucleotide on the parental template strand forms the loop. So both scenarios result in small loops, but on opposite strands.

DNA polymerase "proofreading"

The first line of defense against errors is DNA polymerase proofreading. Both the leading and lagging strand DNA polymerases possess a second enzymatic activity in addition to their primary DNA polymerization activity. It is referred to as a "3'-to-5′ exonuclease activity." 3'-to-5' is the direction opposite that of DNA synthesis.

This ability enables the enzyme to quickly chew back the DNA strand it's synthesizing to immediately remove the error. DNA polymerase proofreading is like the "delete" key on your keyboard. When you make a typo, you hit delete to remove it. Then you keep typing. This is a pretty good analogy for polymerase proofreading.

Here's how it works. When there's a mismatch or an insertion-deletion loop near the growing 3' end of the new strand, it distorts the geometry of the double helix and creates a poor 3' end for the enzyme to extend. The enzyme senses this and moves the problematic 3' end of the new strand from its polymerase active site to its 3'-to-5' exonuclease active site.

The enzyme then literally steps backward and chews off just that one incorrect nucleotide. Once the error is removed, the now normal 3' end shifts back to the enzyme's main polymerization active site and synthesis resumes.

As I mentioned, DNA polymerase proofreading improves fidelity by roughly 100- to 1,000-fold. But this would still leave more errors than the cell can safely tolerate. Thus, human cells employ another error-fixing solution: Mismatch Repair.

Mismatch repair (MMR)

Unlike proofreading, MMR involves multiple steps and multiple enzymes. And whereas proofreading occurs immediately by the polymerase itself. MMR acts soon afterward.

The MMR pathway initiates with detection of the error at or near the replisome. But it often finishes behind the moving replication fork after the polymerase has moved beyond the error.

At this point, I'd like to step back and generalize, because MMR and most of the other repair pathways I'll introduce shortly follow a common logic known as “cut-and-patch repair" which takes advantage of the fact that the double helix carries two separate copies of all the genetic information--one in each of its two strands.

In cut-and-patch repair, a specific protein recognizes an error or lesion. Then another--an endonuclease--cuts that strand in front of the error (5' of the error). Next, an exonuclease begins at that cut and chews up the DNA to remove the error (or lesion). A DNA polymerase and its partner PCNA then replace those nucleotides. Finally, a DNA ligase seals the nick left by the polymerase.

Let’s now focus on MMR, specifically.

The primary MMR detector protein complex--and the initiator of the pathway--is the C-shaped heterodimer MutSα. It recognizes single-base mismatches and small insertion–deletion loops of one or two nucleotides--the kinds of errors we've been discussing.

MutSα diffuses through the nucleus during S phase when the new genome is being synthesized. But MutSα also contains a PIP-box with which it transiently and repeatedly binds to PCNA's PIP binding site. This on-again-off-again binding increases the concentration of MutSα near replication forks--exactly where new errors can be found.

As DNA is synthesized, MutSα repeatedly binds to DNA and probes it for the kinds of abnormalities caused by substitutions, insertions, and deletions. Mismatched base pairs and small insertion-deletion loops make the DNA more flexible than correctly paired DNA. When MutSα encounters one, it grips the DNA and bends, or kinks, it.

Then, a specific amino acid in MutSα wedges between the base pair that includes the mismatched base to test whether it's really an error and also to stabilize the kinked state.

Recognition of the mismatch also promotes ATP binding to MutSα, which causes it to change its shape from its mismatch-recognition C-shape to more like a sliding clamp that encircles the DNA. Now MutSα loosens its grip on the mismatch and slides down the DNA. It also helps recruit and activate other repair proteins, including the next enzyme in the pathway, an endonuclease.

Think of an endonuclease as scissors that cut DNA at an internal site. In MMR, the endonuclease makes one or more cuts--specifically in the newly synthesized strand. It "knows" which is the new strand based on replication-associated signals such as the gaps between Okazaki fragments on the lagging strand, and possibly by the orientation of the PCNA on the leading strand.

At least one endonuclease cut must lie on the 5' side of the mismatch to provide an entry point for the next enzyme, the exonuclease.

Starting at the endonuclease cut, the exonuclease chews up the DNA in the direction of the error--typically hundreds of nucleotides. (Think of Pac-Man in action!) Other repair proteins regulate the exonuclease so that it removes only that limited stretch of the new strand--the stretch that includes the error.

With the error now removed, a DNA polymerase arrives and a new PCNA is loaded right where the exonuclease started chewing. Together, they synthesize new DNA using the parental template strand as a guide, filling the gap. When the polymerase reaches the far end of the gap, the missing DNA will have been replaced. The DNA polymerase and its PCNA then depart.

DNA polymerase can replace the missing nucleotides, but it can't seal the final break in the new strand's backbone. A DNA ligase arrives to do that. It seals the remaining nick, restoring the strand's continuity. The error is now corrected.

Mismatch repair is an amazing capability of the cell. It employs multiple specialized enzymes that must act in a prescribed sequence to detect, remove and then replace the error. In a typical human cell division, MMR might correct tens to a few hundred copying errors that escaped proofreading per genome replication.

In the next chapter, we’ll turn to a different kind of DNA polymerase error: the incorporation of an RNA nucleotide into DNA.